[News] AMD Ryzen 5 1600X Cinebench R15 Score Close To Intel Core i7 6800K

[Warboy]

We start the day with a new Ryzen post. This round a Ryzen 5 1600X performance result in Cinebench R15 surfaced. The result comes from Reddit forum user VulkanDX12.

The man posted a picture of a Cinebench R15 benchmark run showing a CPU score of 1,136. Cinebench lists an AMD Ryzen processor with 6 cores and 12 threads clocked at 3.3GHzand model number ZD3301BBM6IF4_37/33_Y.

Z = QS (Qualification Sample, which is closer to the retail version than an Engineering Sample)
D = Desktop processor (in this case, Summit Ridge)
330 = 3.30GHz base frequency
1 = Model revision number (first revision)
BB = 65W
M = AM4 platform
6 = Number of physical cores
I = ?? (there part indicates the amount of cache, but the decoder doesn't contain an entry for "I")
F4 = B stepping
37 = 3.70GHz boost frequency
33 = 3.30GHz base frequency

If you plot the Ryzen CPU's score in a graph of Cinebench R15 benchmark results over at Anandtech containing scores from 20 Intel processors, its score of 1,136 matches up with (and comes out slightly ahead of) Intel's Core i7-6800K. The second screenshot shows single threaded perf.

Source- Guru3D

[StyM]

competition FTW...
6950X @ 699 ?

[FlanK3r]

No, its impossible fight with 16 threads against 20 threads. Its logic. Ryzen seems solid competition against Haswell-E/Boradwell-E. But (and Im AMD fan) the king will be still 6950X and later SK-X

[dasa]

looks like its single threaded performance is also much the same as 6800k

[Chrono Detector]

These leaked benchmarks look very promising, can't wait to see more really. Hopefully this will force Intel to lower their prices because what they are charging for their top tier CPU's is rather ludicrous due to no competition.

[Warboy]

These leaked benchmarks look very promising, can't wait to see more really. Hopefully this will force Intel to lower their prices because what they are charging for their top tier CPU's is rather ludicrous due to no competition.

Agreed, if the 1600X is on point for the 6800K, then 1700X should be on point for 6900K.

Which means Intel would be more then triple the price.

My only concern is what about memory bandwidth in high memory usage applications but time will tell.

If Intel doesn't lower prices, 389 dollars for the 1700X is very appealing. I wonder what the server-side of Ryzen will be like and if it would work with consumer side boards.

[Metroid]

These leaked benchmarks look very promising, can't wait to see more really. Hopefully this will force Intel to lower their prices because what they are charging for their top tier CPU's is rather ludicrous due to no competition.

It's more than that, charge for overclocking was the worse off all, the best cpu to date where everything got it right was i7 920 and that was 9 years ago, blocking overclock on i3 and most i5 cpus is bad. I have hundreds of computers with Intel boards and CPU's and if the ryzen hype pays off then my next systems will all be with AMD and that will teach Intel something.

[RPGWiZaRD]

Ryzen is starting to look better and better. But we're not stock clock users here and personally I'm thinking I want at least reach like 4.8GHz on a 1700X or otherwise I likely still pick 7700K. But time will tell, if I'd have to make an educated, that's roughly where I'd think it's going to max out on air/water though.

The big question is whether AMD's new overclocking tool which supposedly allow per core clocking (very sweet!), also allows telling which priority order the different cores are to be used. I mean if you can only get Core 0 and 1 to like 4.5GHz but Core 2 and 3 etc goes up to like 4.8GHz on reasonable voltage, that'd suck if you can't get those poorly multithreaded apps to first use the higher clocked cores.

Also I don't see 16 threads vital, I hope we can just use the 8 cores and disable the virtual cores and it'd further optimize the heat output & overclocking capability, 8 cores/threads will be sorta a sweetspot for a long time to come still I'd imagine.

1700X @ ~4.8GHz on 8 cores/threads with IPC ~6800K = sign me up for one!

[BenchZowner]

8 cores @ 4.8GHz... what you sniffin' RPGWiZaRD, it's very good

[sergiu]

Did anyone noticed the HT factor of 1.45? That's huge!

[AliG]

Did anyone noticed the HT factor of 1.45? That's huge!

Doesn't seem that unusual to me; their SMT is very similar to Intel's. Makes sense that threads would scale similarly to the 6800k.

[Andi64]

Ryzen is starting to look better and better. But we're not stock clock users here and personally I'm thinking I want at least reach like 4.8GHz on a 1700X or otherwise I likely still pick 7700K. But time will tell, if I'd have to make an educated, that's roughly where I'd think it's going to max out on air/water though.

The big question is whether AMD's new overclocking tool which supposedly allow per core clocking (very sweet!), also allows telling which priority order the different cores are to be used. I mean if you can only get Core 0 and 1 to like 4.5GHz but Core 2 and 3 etc goes up to like 4.8GHz on reasonable voltage, that'd suck if you can't get those poorly multithreaded apps to first use the higher clocked cores.

Also I don't see 16 threads vital, I hope we can just use the 8 cores and disable the virtual cores and it'd further optimize the heat output & overclocking capability, 8 cores/threads will be sorta a sweetspot for a long time to come still I'd imagine.

1700X @ ~4.8GHz on 8 cores/threads with IPC ~6800K = sign me up for one!

So, you are expecting "at least" 4800Mhz on a 8c/16t CPU when Intel can barely hit 4500Mhz with a 6c/12t CPU...

[Sparky]

I'm just not going to get hyped up, honestly. I really really hope we have a winner here and the "old" AMD is back to give Intel a good swift kick like in the good old days, but, I'm trying to be cautious also.

[RPGWiZaRD]

So, you are expecting "at least" 4800Mhz on a 8c/16t CPU when Intel can barely hit 4500Mhz with a 6c/12t CPU...

No I'm just stating what I want to see before I'm going AMD since a lot of applications I use still have poor multithread useage so even if it has core advantage, I don't want the lesser threaded application performance scenarios fall too much behind (10~15% max is where my ego can handle :p).

But I said 4.8GHz with 8 cores and HT turned off. Also let's remember AMD has per core clocking ability now, before the least stable core would be the decisive factor, having per core clockability improves the overclocking ability and depending how you count it I'd guess the average core frequency could be seen as the "traditional" frequency ratio in this case, in the past if a core would only make it to 4.4GHz but another could have made it to 4.6GHz you still had a 4.4GHz clock. With AMD's solution if we'd agree average core clock between cores would be the "rounded down" frequency, that would definitely help, my expectation is 100~200MHz higher clock as a result which also adds up.

Of course 4.8GHz is a high count for a 8 core CPU that I don't disagree with but I don't see it impossible with AMD's supposedly very good energy efficiency tweaks. But anyway it's just a bet guys, nothing more, I'm the guy who likes betting pre-release and see how close I get.

So 4.8GHz with HT off and average frequency between cores rounded down was my bet, with HT on I'd say avg 4.6~4.7GHz (meaning worst samples 4.4GHz and good sample 4.8GHz'ish). This is on good water/air and reasonable high 24/7 volts.

[Andi64]

I really hope you are right. I'm with you on the 10-15% single core performance loss. If Ryzen 8C/16T is ~10% behind Intel in single threaded applications I'm going back to AMD.

But I don't think they'll clock that high. I'd bet ~4.3/4.4Ghz average (all cores), even on 4C/4T CPUs. It's a new architecture on a new manufacturing process for GF. I dont want to have high expectations. We'll see.

[AliG]

The process really isn't all that new. As I understand it, they bought a lot of the technology from IBM.

[dasa]

Doesn't seem that unusual to me; their SMT is very similar to Intel's. Makes sense that threads would scale similarly to the 6800k.

I didn't know ht scaled better on 6800k than 7700k ~30%
Cant find much in the way of tests done on it

[AliG]

I didn't know ht scaled better on 6800k than 7700k ~30%
Cant find much in the way of tests done on it

Probably has more to do with the test than the chips being different. The only real advantage the 6800k's cores have is the larger l3 cache.

[chew*]

I think people want 4.8-5.0 chips but the reality is we are getting to the point where higher base clock chips are harder to hit targets for manufacturers.

I recall reading some article about some electrical moore's law as we go lower in nm....

It basically states right about where we are now...companies reach a clock threshhold (wall) so at that point they will start adding more cores. And here we are today. 16 to 20 threads will be avg...

[Sparky]

Yeah only so fast and so tiny you can get silicon to go before you just run into physical limitations of the material.

[qcmadness]

The process really isn't all that new. As I understand it, they bought a lot of the technology from IBM.

Should be the same process as Polaris 10 and 11 are using.

[AbortRetryFail?]

...

I recall reading some article about some electrical moore's law as we go lower in nm....

...

Pretty much what Sparky said.

Silicon nuclei are roughly 5nm. What has been found below 10nm is that the nuclei struggle to carry electric current because they are so narrowly compacted in the "gate" (as in, "7nm").

Enter silicon-germanium (SiGe) at 7nm, which improves electron mobility and allows higher currents by "exciting" the silicon nuclei.

My understanding is 10+ years ago IBM/AMD were using SiGe in a limited fashion, and a few years ago IBM demonstrated working gates at 7nm using it. I'll guess that this IP was part of the sale of the Big Blue fab'ing to GloFo. And apparently now in 2017 ... there are other elements being combined into silicon alloys for the same purpose.

As far as 4.8GHz, I'll take a swing at it: A 95w R5 (4/8) might push it with good air --- at $200 or less?

[AliG]

As far as 4.8GHz, I'll take a swing at it: A 95w R5 (4/8) might push it with good air --- at $200 or less?

I think that depends on whether we're looking at true 4-core or a cutdown 8-core. I can see a native quadcore clocking higher than a native octocore, but a cutdown chip implies that couldn't stably run at those frequencies within the power targets. If that's the case, I doubt you'd actually see much of an increase in overclocking headroom (despite the lesser heat output).

[vario]

I think that depends on whether we're looking at true 4-core or a cutdown 8-core. I can see a native quadcore clocking higher than a native octocore, but a cutdown chip implies that couldn't stably run at those frequencies within the power targets. If that's the case, I doubt you'd actually see much of an increase in overclocking headroom (despite the lesser heat output).

Well, i think that a simple reality is, if AMD COULD clock zen chips higher, they would, especially the cutdown 6 and 4 core ones, because in this state they dont really have a direct head to head intel competition for the gaming market.
They have what they have and doing their best with it.But looking at all those cpus , they all pretty much are clocked almost the same, which simply implies they cant reach higher or at best cant do it repetitively.

[chew*]

The main thing that is often overlooked is ipc per core per clock.

So we will need to pay close attention to that.

If ipc of amds 4.0 beats haswell 4.2 then reality is you only need 4.3 to be equal to 4.5.

Saying you expect Xx speeds is pointless without knowing this factor.

The analogy of dropping cores or ht is plausible to an extent.

Example 1.3v gets me 4.5 ht on or 5.0 ht off but we are talking a 4 core chip. That result with 8 cores ht on would likely be halved...making gains from threading on and off less drastic and not as beneficial.